SYNTACTIC LYNKAGE

CHAPTER II

DISCUSSION

A.    INTRODUCTION

Syntactic linkage has to do with the devices (mainly morphological) by which speakers can signal which words, phrases or clauses are linked. It is a general concept which subsumes the traditional concepts of agreement and government. As we will see, the traditional distinction does not sit easily with the analysis of phrases into heads and modifiers, with heads controlling the modifiers. Some parts of the traditional concepts are compatible with the view that verbs are the heads of clauses; but others are not, particularly the relationship between subject noun and verb.

This chapter contains data from languages other than English. The book deals essentially with syntax for students in English (Language) courses, and most of the examples are from English. The problem is that English clauses have minimal syntactic linkage compared with, say, Latin or Russian (not to mention the marvellously intricate systems of linkage to be found in Bantu languages, in many Native American languages and in Australian languages). Agreement and government may not be prominent in Modern English, but they are central concepts of syntax and are prominent in the syntax of many languages, including Early English. For these reasons, syntactic linkage must be included, but examples are taken from other languages as well as from English.

B.     AGREEMENT

Agreement is found inside noun phrases and, in some of the Indo- European languages of Europe, in the copula constructions – adjective complements of Be (and other copula verbs) agree with the subject noun. Agreement in noun phrases is exemplified in (1).

(1)   Latin

a.       magnus reks

great king ‘a/the great king’

b.      magna regina

great queen ‘a/the great queen’

c.       magnum oppidum

great town ‘a/the great town’

The phrases in (1a–c) could function as subjects of clauses. Reks in (1a), regina in (1b) and oppidum in (1c) are said to be in the nominative case, the case being signalled by the suffixes -s, -a and -um. The adjective translated as ‘great’ consists of the stem magn- [as in ‘magnify’ to make great] plus various suffixes, -us, -a and -um. Which suffix is added is controlled by the head noun. *Magna reks is incorrect, as is *magnus regina. The nouns in (1a–c) have singular number; different suffixes appear both on the nouns and the adjectives when they are plural, as in (2).

(2)   Latin

a.        magni reges

great kings ‘(the) great kings’

b.      magnae reginae

great queens ‘(the) great queens’

c.       magna oppida

great towns ‘(the) great towns’

The nouns in (2) are still in the nominative case and can still function as the subjects of clauses, but they are plural and have different suffixes, and the adjectives too have different suffixes. Whether, for example, the stem magn- takes the suffixes -us, -a, -i or -ae depends on what type of noun it modifies, what case the noun is in and whether the noun is singular or plural. That is, the noun is the controlling word but both noun and adjective change shape, that is, change their suffixes; this is why the term ‘agreement’ is used. The traditional formula is that adjectives agree with nouns in number and case (and also in gender, which is discussed below.) In spite of the traditional formula, we can view the noun as governing the adjective in case and number.

 Old English had similar patterns of agreement between the head noun in a noun phrase and adjectives that modified it. Consider the examples in (3).

(3)   Old English

a.       go¯d cyning (good king)

b.      go¯du cwe¯n (good queen)

c.       go¯d scip (good ship)

The nouns in (3) are in the nominative case. Cyning in (3a) is masculine,and scip in (3c) is neuter. The adjective in both examples consists of thestem go￣d, with no suffix. Cwe￣n in (3b) is feminine, and the adjective takesthe suffix -u, go￣du. A richer pattern is observed when the head noun isplural, as in (4).

a.       go¯de cyningas (good kings)

b.      go¯da cwe¯na (good queens)

c.       go¯du scipu (good ships) 

The masculine plural noun cyningas in (4a) is modified by go￣de with the suffix -e. The feminine plural noun cwe￣na in (4b) is modified by go￣da, with the suffix -a. Finally, in (4c) the neuter plural noun scipu is modified by go￣du, with the suffix -u. Although the adjective and noun in (3a) and (3b) do not change their shape, since no suffixes are added to the stem, all the nouns and adjectives in (4) do change their shape, since suffixes are added to both adjective and noun. Looking at the overall pattern for singular and plural, and for other cases, we can justifiably speak of adjective and noun agreeing in Old English.

C.    GOVERNMENT

Consider now the examples in (5).

(4)   Latin

a.       Servus              magnum          regem occidit

slave                great                king     killed

‘A/the slave killed a/the great king’

b.      Servus              magnos           reges    occidit

slave                great               kings    killed

‘A/the slave killed (the) great kings’

c.       Servus  magnam          reginam           occidit

slave    great               queen               killed

‘A/the slave killed a/the great queen’

Regem in (5a) is the direct object of occidit and has the suffix -em. The adjective has the suffix -um. In (5b) reges is the direct object of occidit, is plural and has the suffix -es. The adjective has the suffix -os. In (5c), reginam is the direct object of occidit, is singular and has the suffix -am. The adjective has the same suffix. Regem, reginam and reges are said to be in the accusative case. Note that reges, with the same suffix -es, is subject in (2a) and direct object in (5b), but that the adjective has different suffixes.

Occidit is a verb that requires an object in what is called the accusative case. In the examples in (5), occidit, in the traditional formula, governs its object noun in the accusative case; that is, it assigns accusative case to the stems reg- (king) and regin- (queen). Independently of the verb, these nouns are singular in (5a) and (5c) and plural in (5b). The combination of accusative and singular requires the choice of the suffixes -em or –am depending on the stem, and the choice of accusative and plural requires -es for the stem reg-. The properties ‘accusative’ and ‘singular’ or ‘accusative and plural’ are passed on to the adjective in the direct object noun phrase, magn-, and the appropriate suffix is chosen.

The majority of verbs in Latin assign accusative case to their object noun, but many verbs assign one or other of the remaining three cases. For example, the verb utor (I use) governs its object noun in the ablative case, as shown in (6).

a.       Reks          gladium           ponit

King          sword                         is-putting-down

‘The king is putting down the sword’

b.      Reks          gladio             utitur

king           sword              is-using

‘The king is using the sword’

For verb and object noun too, a similar pattern occurs in Old English, as demonstrated in (7).

a.       se eorl slo¯g go¯dne cyning

that warrior killed (a)-good king

b.      se eorl slo¯g gode cwe¯ne

that warrior killed (a)-good queen

In (7b), the object noun cwe￣ne, with the suffix -e, is modified by the adjective go￣de, with the suffixe -e. This contrasts with the subject noun cwe￣n in (3b) with no suffix and modified by the adjective go￣du, with the suffix -u. In (7a), the noun cyning has no suffix but the adjective does have a special suffix, namely -ne as in go￣dne. There is no contrast in the suffixes added to subject and object plural nouns.

The patterns of suffixes are summed up by saying that slo￣g requires its object noun to be in the accusative case. Other verbs require their noun to be in a different case; andwyrdan ‘answer’, for instance, requires its object noun to take dative suffixes, as shown in (8).

(8) se eorl andwyrde go¯dum cyninge

that warrior answered (a)-good king

Cyninge in (8) has the dative case suffix -e and go￣dum has the dative suffix -um. The verb bı￣dan ‘wait for’ requires its object noun to take a genitive suffix, as in (9).

(9) se eorl ba¯d go¯des cyninges

that warrior waited-for (a)-good king

In (9), cyninges has the genitive suffix -es and the adjective go￣des has the same suffix.

Prepositions in Latin also assign case to their complement nouns. Ad (to) governs nouns in the accusative case, and de (from) governs nouns in the ablative case.

The nominative case was thought of as the case that was used when speakers were using nouns to name entities. The theory was, and indeed still is, that speakers pick out and name an entity and then say something about it. (See Chapter 8 on grammatical functions.) ‘Accusative’ looks as though it should have something to do with accusing; this does not make much sense but results from a mistranslation into Latin of a Greek term meaning ‘what is effected or brought about’. It seems that the central examples of accusative case were taken to be the Classical Greek equivalents of ‘She built a house’. ‘Ablative’ derives from a Latin word meaning ‘taking away’ and occurs with prepositions expressing movement from or off something, as in (10b).

a.        ad hortum

to garden   ‘to the garden’

b.      e horto

out-of garden        ‘out of the garden’

The above Latin examples show that different case-number suffixes are added to nouns and adjectives and that which suffix is chosen in a particular clause depends on the noun. Latin nouns fall into a number of different classes, known as genders. Gender will not be discussed here but will be examined in Chapter 12 on grammar and semantics.

Prepositions in Old English also require their complement nouns to have particular case suffixes. For example, to￣ ‘to’ governs its complement noun in the dative case (like most prepositions), but †urh ‘by means of ’ governs its complement noun in the accusative case. The distinction is exemplified in (11).

a.        to¯ tæ¯m cyninge (to that king)

b.      turh tone cyning (by-means-of that king)

In (11a), †a￣m is the dative form of se ‘that’ and cyninge has the dative suffix -e. In (11b), †one is the accusative form of se but cyning has no suffix.

D.    NUMBER AND PERSON LINKAGE

We turn now to the final strand of syntactic linkage in Latin, the relationship between subject nouns and verbs. Consider (12).

(12) Latin

a. reks legit (the-king is-reading)

b. reges legunt (the-kings are-reading)

In (12a), reks is singular and the verb leg- (read) has the suffix -it, which is singular. In (12b), reges is plural and leg- has the suffix -unt, which is plural. The traditional formula is that the verb agrees with the subject noun in number (and person). This analysis accords with the view that the subject noun (phrase) is more important than the other nouns in a clause because speakers use subject nouns to name the entity they want to talk about. At various places in this introduction, however, we have seen that there are good reasons to consider the verb the head of a given clause, controlling all the other words and phrases in it. From the perspective of dependency relations (Chapter 1) and the lexicon (Chapter 5), the relation between the verb and subject noun is no different from the relations between the verb and other nouns in clauses.

Whichever analysis readers favour, the term ‘agreement in number’ is not accurate. Either the subject noun imposes a given number on the verb or the verb imposes a given number of the noun. That is, both analyses recognise a controlling word, and where there is a controlling word we are dealing with government.

Verbs in Latin signal more than just number. Other suffixes are added to verb stems in Latin, as shown in (13).

            (13) Latin

a)      epistolam lego

letter I-am-reading ‘I am reading the/a letter’

b)      epistolam legis

letter you (sg)-are-reading

‘You (singular) are reading the/a letter’

c)      epistolam legimus

letter we-are-reading ‘We are reading the/a letter’

The contrast between lego in (13a) and legis in (13b) has to do with who is presented as reading, the speaker or the addressee. The speaker is considered the central participant in a conversation, the first person. The addressee is the second in importance, that is, is the second person. The speaker and others can be presented as jointly doing something, as in (13c). Lego is described as being in the first person singular, legimus as being in the first person plural and legis as being in the second person singular. In (12a, b), the speaker presents the action as being performed by neither the speaker nor the hearer but by a third person in (12a) and by third persons in (12b). Legit is said to be in the third person singular, and legunt is described as being in the third person plural. Latin has pronouns, but they were not used unless for emphasis.

E.     SYNTACTIC LINKAGE IN ENGLISH

Syntactic linkage in English is less complex than in Latin for the simple reason that English nouns no longer have a case system like the Latin one and English verbs have only one contrast in person and number. English uses word order and prepositions to signal syntactic links. As discussed in Chapter 8 on grammatical functions, in the active, declarative construction the subject and object nouns are never preceded by prepositions, the subject noun is immediately to the left of the verb (allowing for modifiers such as relative clauses) and the direct object noun is immediately to the right of the verb. Adjectives modifying a given noun are immediately to its left. Compare the word order in the line of Latin poetry in (14) and the word order in the English translation.

(14) Latin

Cerberus et nullas hodie petat improbus umbras Cerberus and no today let-him-seek voracious shadows ‘Let voracious Cerberus seek no shadows today’ (Propertius). The adjective improbus, the second last word in the clause, modifies the first word, Cerberus, and the adjective nullas, the third word in the clause, modifies the last word, umbras. (Note that (14) is a line of poetry; native speakers of Latin, as far as we can make out, did not talk off the cuff in this complex fashion.) In the English translation, voracious immediately precedes Cerberus and no immediately precedes shadows. English pronouns have retained some case distinctions, and these are relevant to syntactic linkage. In standard (written) English, I, we, he, she and they occur as subject and me, us, him, her and them occur as direct, indirect or oblique object. A couple of examples are given in (15).

a.       He told her the whole story.

b.      She told him the whole story.

The construction in (16) – called the [CONTRAST] construction in Chapter 3 – does have an unusual word order in that the direct object or oblique object phrase is put at the front of the clause.

a.       The other plan she rejected out of hand.

b.      To her eldest son she left the Aubusson tapestries.

The entire object phrase is moved to the front of the clause, that is, the noun and all its modifiers. The noun and its modifiers are next to each other but the whole noun phrase is in an unusual position in the clause. Relative clauses and noun complement clauses (see Chapters 6 and 7 on clauses) can and do occur separated from their head noun, as in (17).

a.       The vase got broken that Sheila had brought all the way from Chengdu.

b.      The plan was rejected out of hand that traffic should be banned.

c.       The idea dismayed the Prime Minister that the Dome was dull.

In (17a), the relative clause that Sheila had brought all the way from Chengdu is separated from its head noun vase by the phrase got broken. The vase got broken is an intransitive construction and there is no other noun that might be taken for the head modified by the relative clause. Noun complement clauses can be separated from the noun they modify even in transitive clauses. Example (17b) poses no problem because it is intransitive and the noun complement clause that traffic should be banned can only be interpreted as modifying plan. In (17c), which reads much more awkwardly than (17b) but is nonetheless acceptable, the noun complement clause that the Dome was dull relates to the content of an idea, plan, proposal, theory and so on and can only be interpreted as modifying idea. (Noun complement clauses always relate to the content of an idea and so on; this is one of the criteria by which you can decide whether a given clause is a noun complement or a relative clause.) In general, then, phrases in English can be separated from the word they modify and so can clauses, but modifier words are not separated from their head word.

 English has a minimal amount of linkage between subject noun and verb. Only the verb Be has any linkage in the past tense, was requiring a singular noun and were requiring a plural noun (in standard English). All other verbs have one past-tense form (and are said to be invariable in the past tense). In the present tense, Be is again idiosyncratic; it has a first person singular form am in addition to the contrast between is (third person, singular) and are (second person singular and all persons plural). Apart from the modal verbs such as can, must and so on, which are invariable, all other verbs take the suffix -s in the third person singular – The dog barks vs The dogs bark.

F.     NUMBER IN ENGLISH

The person-number relationship between subject verb and noun in English is simple compared with the large number of suffixes in Latin, yet the English system is not straightforward. One complexity is that while we are here focusing on the verb system in standard English, most of the population of the UK speak some non-standard variety. Many nonstandard varieties differ from the standard one; in some, speakers have generalised the -s suffix to all persons – We sits quietly in the classroom, you knows what I’m talking about, while in others the speakers have removed the -s suffix – She live in that street.

 Even speakers of standard English itself disagree over whether a singular or plural verb should be used with certain nouns. For instance, committee is a singular noun but is used to refer to groups of more than one person. Many speakers and writers use either a singular or a plural verb depending on whether they view a particular committee as several people or as a single unit. The committee are discussing this proposal today presents the committee as several individuals who have to discuss the matter and work out a decision. The committee is discussing the proposal today presents the committee in a more impersonal manner, as a single unit. Every generation, however, appears to contain speakers who believe that if a noun is singular the verb should always be singular. Such speakers are becoming rarer.

G.    GENDER IN ENGLISH

Nouns in English do not fall into different grammatical classes of the sort found in Latin. As we have seen in the discussion in Chapter 5 on the lexicon, English nouns do split up into classes, but the split is not unrelated to the meaning and is not reflected in syntactic linkage. Consider the examples in (18).

a.       My sister phoned. She said that the children were back at school.

b.      Your brother went out. He said he was going to the bank.

c.       The car is ready. It just needed new points.

The pronoun she is used in reference to a female, he in reference to a male and it in reference to an entity that is neither male nor female. In contrast, there is no obvious reason why the Latin noun mensa (table) is said to have feminine gender, nor why hortus (garden) is labelled ‘masculine’ and oppidum (town) is labelled ‘neuter’. Despite the simplicity of the (typical) examples in (18), English usage is not without its subtleties. It is said that some speakers of English refer to cars and boats as though they were female, as in She’s a wonderful car, but this usage appears to be falling into disuse.

Reference to babies and animals varies. Parents of a baby and owners of, in particular, cats and dogs, know the sex of their child or animal and use he or she. Speakers who do not know the identity of a particular baby often refer to the baby by it (indeed I was about to write ‘often refer to it’). The choice of it is usually caused by lack of knowledge; once particular babies have been introduced as Selena or Rachel, Angus or Torquil, they will thereafter be referred to by he or she as appropriate. But just to demonstrate that context is important, parents can be observed referring to their baby by it, a usage that typically expresses love for the baby. Similarly, the choice of pronouns for animals is affected by the use of proper names. Example (19a) is perfectly normal, (19b) is peculiar.

a.       The dog slunk under the table. It had eaten the sandwiches.

b.      Towser slunk under the table. He (??it) had eaten the sandwiches.

We close this chapter with a reminder that the relations implicated in syntactic linkage are dependency relations holding between heads and modifiers. Some relations cross clause boundaries (especially the boundaries of relative clauses), but mostly they hold within single clauses.

CHAPTER III

CONCLUTION

Many languages have devices which signal which words belong together as head and modifiers; that is, they signal syntactic linkage. Traditionally, two types of devices are recognised, namely agreement and government. Agreement was applied to examples in which a head and its modifier were both marked for some property. In many languages, the head noun in a noun phrase and the adjectives that modify it all carry markings for case and number, while the subject noun in a clause and the verb both carry markings for person and number. Government was applied in constructions in which the head was not marked but the modifier was. For example, in many languages the direct object of a verb carries a particular case suffix but the verb itself has no marking. Likewise, in prepositional phrases the complement of the preposition has a case suffix but the preposition itself has no marking. In spite of the traditional distinction, instances of agreement and government all involve a head assigning some property or properties to its modifiers. The current preferred view is that all syntactic linkage is government.